JP3479024B2 - Electron beam exposure mask and pattern design method thereof - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electron beam exposure mask used for transferring a fine pattern in a semiconductor device manufacturing process and a pattern design method therefor, and more particularly to an electron beam for preventing deformation of a stencil mask. The present invention relates to an exposure mask and its pattern design method.

[0002]

2. Description of the Related Art In a manufacturing process of a semiconductor integrated circuit device, a microfabrication technique for drawing an integrated circuit pattern using electron beam (EB) drawing and a focused beam of a charged particle beam such as an ion beam is put into practical use. Has been done. For example, in an electron beam exposure apparatus, a wafer coated with an electron beam resist is irradiated with an electron beam to directly expose an integrated circuit pattern onto the electron beam resist, and an EB mask is used to obtain a drawing pattern by the electron beam. Has been done. As an electron beam drawing technique using such an electron beam, there is a partial collective exposure method in which a mask pattern is reduced and projected to collectively draw a unit area such as a memory cell. Further, as a mask used for collective exposure for collectively irradiating a circuit pattern on a wafer, for example, there is a stencil mask having a hole through which an electron beam passes to transfer the circuit pattern.

However, in a stencil mask, a region that is entirely surrounded by mask holes does not have a portion that supports this region, so that a stencil mask having such a pattern cannot be manufactured. Such a problem is generally called a donut problem.

Therefore, conventionally, for a mask pattern having a donut problem, a stencil mask is manufactured by using two masks. Figure 10
(A) is a plan view showing a designed mask pattern having a donut problem, (b) is a plan view showing an example of a complementary mask to the mask pattern shown in (a), and (c) is another example of a complementary mask. It is a top view.

For example, when a mask pattern 20 composed of a donut pattern 20a and a line pattern opening 20b provided therein as shown in FIG. 10 (a) is designed, as shown in FIG. 10 (b). , A first complementary mask 21 having a half mask hole 23a of the donut pattern 20a and a mask hole 23b of the line pattern opening 20b, and a second half mask hole 23c of the remaining half of the donut pattern 20a. Complementary mask 22
Is used. Further, as shown in FIG. 10C, the first complementary mask 21 and the donut pattern 20a in which the mask hole 23d corresponding to one side of the donut pattern 20a and the mask hole 23b of the line pattern opening 20b are formed. The second complementary mask 22 having the mask holes 23e corresponding to the remaining three sides may be used.

Thus, the two complementary masks 21 and 2 are
By using 2, the inner region of the donut pattern 21a is supported and the designed line pattern can be left in that region.

[0007]

However, by using the complementary masks 21 and 22 as described above,
Although the stencil mask can be manufactured while avoiding the donut problem, there is a problem in that the manufactured stencil mask is deformed such as warped. That is,
For example, in the second complementary mask 22 shown in FIG. 10B, the tongue-shaped region is supported with a point on the line segment connecting the ends of the mask hole 23c as a fulcrum, but in the boundary region. If the strength is low, the region extending in a tongue shape may be deformed, such as warped, and if the deformation is large, damage may occur. Such a modification is the second modification shown in FIG.
It becomes remarkable with the complementary mask of. In addition, FIG.
The same may occur in the first complementary mask 21 shown in FIG. Such a problem is sometimes called a ton problem. Also, the edges of the mask pattern are close to each other,
When the portion supporting the tongue-extending portion is extremely small, the above-mentioned deformation is remarkable. Such a problem is sometimes called a leaf problem.

Further, in a line and space (L / S) pattern such as a bit line and a word line of a memory, there is a problem of deformation of a stencil mask. Figure 11
(A) is a plan view showing the stencil mask immediately after forming the mask holes, and (b) is a plan view showing the stencil mask after a cleaning step after forming the mask holes.

As shown in FIG. 11B, in the stencil mask 24 for the L / S pattern, in the cleaning process performed after forming the mask holes 25, adjacent line portions are caused by surface tension of pure water or the like. There is a problem that they come into contact with each other and cannot be separated due to intermolecular force.

The problem of such contact between line portions exists not only in the L / S pattern but also in various patterns as long as the deformation is larger than a certain level.
Considering the increasing number of complicated mask patterns, it is extremely difficult to manufacture the stencil mask at present.

The present invention has been made in view of the above problems, and is a mask for electron beam exposure which is less likely to be deformed even by a cleaning process after forming a mask hole and can perform exposure with high dimensional accuracy. It is an object to provide a pattern designing method.

[0012]

A first electron beam exposure mask according to the present invention has one grid pattern between adjacent grid pattern regions when divided into a plurality of grid pattern regions having a constant size. The opening exists only in the area, and the opening exists
The grid-like regions are characterized by a checkerboard pattern .

A second electron beam exposure mask according to the present invention has a plurality of grid-shaped regions partitioned in a grid pattern.
The first selection area that is not adjacent to the
At least one first opening, and the grid-like region
At least one of the areas other than the first selection area
Is also divided into one area and is smaller than the grid area
One of a plurality of second selection areas and the second selection area
Of the second opening formed in the
An area other than the first selection area and the second selection area,
And a third selection region in which an opening is not formed .

In the present invention, the pattern of the complementary mask is conventionally determined only for avoiding the donut problem, whereas an opening exists only in one lattice-shaped region between the lattice-shaped regions adjacent to each other. Or, since there is no opening in the grid-shaped region where the first and second openings do not exist, there is no portion where the deformation locally increases,
For example, the mask is unlikely to be deformed even in the cleaning step with pure water. As a result, the yield of the electron beam exposure mask is increased and the dimensional accuracy is improved.

The fixed size can be determined in association with the Young's modulus and thickness of the material and the mask pattern to be exposed.

[0016]

[0017]

A pattern designing method for an electron beam exposure mask according to the present invention is the pattern designing method for an electron beam exposure mask, which comprises a step of assigning a mask pattern to a plurality of complementary masks. step displacement caused by the stress acting on each position in the mask have a step of determining the shape of the opening to be less than the predetermined value, determines the shape of the opening, the mask
The process of partitioning the pattern into a plurality of grid-like regions of a certain size.
And adjacent grid-shaped regions with different complementary masses.
And a step of allocating the same to each other.

[0019]

The step of partitioning the mask pattern may include the step of determining the Young's modulus and thickness of the material of the complementary mask and the dimension of the grid-like region in association with the mask pattern.

Further, the mask pattern is a line-and-space pattern, and the step of determining the shape of the opening can be performed in association with the width, length and thickness of the line portion. .

[0022]

BEST MODE FOR CARRYING OUT THE INVENTION A pattern design method for an electron beam exposure mask according to an embodiment of the present invention will be specifically described below with reference to the accompanying drawings. The first embodiment is shown in FIG.
This is a pattern design method for the mask pattern shown in 0 (a). FIG. 1 is a flowchart showing a pattern design method according to the first embodiment of the present invention.

In the first embodiment, first, when the mask pattern shown in FIG. 10A is divided into checkered grid-like regions and each region is formed into two complementary masks, A checkerboard-like lattice size such that the displacement due to the stress acting on each position is less than a predetermined value is determined based on the mask material to be used, the thickness, and the pattern size (step S1). At this time, there is no particular need to consider whether or not there is a donut pattern in the checkered grid area.

For example, in a portion which becomes a supporting beam at both ends of the L / S pattern or the like, the maximum displacement δ ₁ obtained by the following formula 1 is less than half the width of the opening, and in a portion which becomes a cantilever beam such as a tongue pattern. The maximum displacement δ ₂ obtained by the following equation 2 is set to be less than half the width of the opening.

[0025]

[Equation 1]

[0026]

[Equation 2]

In Equations 1 and 2, P is the size of the uniformly distributed load, a is the length of the beam, E is the Young's modulus of the mask material, I is the second moment of area of the beam, and W is the unit of the beam. It is the weight per length. Also, the second moment of area I
Is expressed by Equation 3 below, where b is the width of the beam and t is the thickness of the beam.

[0028]

[Equation 3]

The setting standard for the predetermined value of displacement in the present invention is not limited to these equations, and setting criteria based on various empirical equations and the like can be provided. For example, the surface tension of pure water used for cleaning can be considered.

Next, the mask pattern is divided into the lattice dimensions determined in the above process (step S2).

Next, it is determined whether or not a donut pattern is present inside each area divided into a checkerboard-like lattice (step S3).

As a result of this judgment, if there is no donut pattern in any grid-shaped area, at that time, the patterns in the grid-shaped areas which are not adjacent to each other in the grid-shaped areas divided in a checkered pattern are gathered. The collected and collected pieces are determined as mask patterns of two complementary masks (step S4).

On the other hand, if the result of the determination is that there is a grid-like region in which the donut pattern exists, the steps from the determination of the grid size to the determination of the donut pattern existence are performed for that region, and the donut pattern exists in the determination. These steps are repeated until the checkerboard lattice size is made smaller than the previous value, for example, to half the previous lattice size, until it is no longer necessary (steps S1 to S).
3).

Then, when it is determined that the donut pattern does not exist in the lattice-shaped region having the minimum size,
A mask pattern to be set for the two complementary masks is determined (step S4). That is, with respect to the grid-like region where the donut pattern does not exist in the result of the first determination,
As it is, the patterns in the grid-like areas that are not adjacent to each other are gathered together in the grid-like area that is divided into the checkered pattern, and for the grid-like area where the donut pattern is present,
Patterns that are sequentially partitioned and are not adjacent to each other are collected together, and the finally collected patterns are determined as mask patterns of two complementary masks.

FIGS. 2A and 2B are plan views showing the first and second complementary masks obtained by the first embodiment, respectively. In the first embodiment, since the lattice size is determined so that the displacement due to the stress acting on each position in the mask is less than the predetermined value, as shown in FIGS. 2 (a) and 2 (b), Neither the Tan problem nor the Leaf problem exists in both the first and second complementary masks 1 and 2.
In addition, since it is divided into a checkerboard-like grid area, the donut problem of the grid size or more is resolved at that point, and even if a donut pattern exists in the grid area, the presence or absence of the donut pattern can be determined after that. All the donut problems have been resolved as we go and repartition.

As shown in FIGS. 2 (a) and 2 (b), the density of the mask holes 3 is equal to that of the first and second complementary masks 1, 2.
2 is almost uniform, and when electron beam exposure is performed using the first and second complementary masks 1 and 2, since the beam drips are almost the same, the Coulomb blur becomes almost the same, and High dimensional accuracy can be obtained for the exposure pattern.

The above-mentioned method is a method of dividing a pattern and allocating it to two complementary masks. However, when the pattern density is high, the Coulomb effect is taken into consideration and the mask pattern is divided into three or more pieces to make three pieces. It may be assigned to the above complementary masks.

In many mask patterns,
By dividing into checkered grid-like regions as in the first embodiment, the density of mask holes between complementary masks, that is, the pattern density, is substantially uniform, for example, the deviation is within 10%, but, for example, 3 When a complementary mask is used, the pattern density may be biased by simply dividing the pattern into checkered patterns in consideration of the distribution of displacement due to the mask material or the like. In such a case, in order to equalize the Coulomb blur sizes, it is desirable to consider that the deviation of the pattern density is within 10% when determining the lattice size.

Further, in the first embodiment, when a donut pattern exists in the grid-shaped area, the determination is made again by dividing the grid into a checkered grid-shaped area. If the displacement due to the stress acting on the mask is less than a predetermined value and the tongue problem does not occur even if the donut pattern in the inside is divided into halves, simply dividing the donut pattern into halves as in the conventional case. Good.

Next, a second embodiment of the present invention will be described. FIG. 3 is a flowchart showing a pattern design method according to the second embodiment of the present invention.

In the second embodiment, first, the mask pattern is divided into a checkerboard pattern having an appropriate lattice size based on experiments or experience without considering the mask material, thickness, etc. (step S11). This grid size is, for example, 1
The thickness is about 0 to 40 μm, but is not limited to this.

Next, for each grid-like region, the pattern arranged in that region is extracted, and the displacement due to the stress acting on each position is less than a predetermined value, and whether or not there is a donut pattern. Is determined (step S1
2).

As a result of this determination, if there is no position where the displacement is equal to or greater than the predetermined value in any area and there is no donut pattern, at that time, the areas are adjacent to each other in a checkered grid area. The patterns in the non-lattice region are gathered together, and the gathered patterns are determined as the mask patterns of the two complementary masks, respectively (step S
13).

On the other hand, as a result of the judgment, when there is a position where the displacement is equal to or more than a predetermined value or when there is a donut pattern, such a grid-like region is divided into sections having an appropriate size as described above, for example. A determination process is performed, and these processes are repeated until the stress at each position becomes less than a predetermined value and the donut pattern disappears (steps S11 and S12). The size of the lattice size here may be smaller than that in the previous stage, and is, for example, ½ of the lattice size in the previous stage.

Then, when it is determined that the displacement at each position in the lattice-shaped region having the minimum size is less than the predetermined value and that the donut pattern does not exist, the mask patterns set in the two complementary masks are obtained. Is determined (step S13).

Also in the complementary mask thus obtained, the donut problem, the tongue problem and the leaf problem are solved as in the first embodiment. Such a method
This is effective when the pattern to be produced is not clear. Further, in the second embodiment, if the grid size is determined to be small at this time, the displacement becomes less than the predetermined value in the subsequent determination, and it is often considered that the donut pattern does not exist. On the other hand,
Since the material and thickness of the mask are not taken into consideration when determining the lattice size, the number of steps is remarkably reduced depending on the mask pattern by determining the lattice size to be small, for example, about 5 to 20 μm. be able to.

According to the second embodiment as well, in many mask patterns, the pattern density is automatically almost uniform, for example, the difference is within 10%, but the pattern density is biased depending on the mask pattern. It may happen. In such a case, in order to make Coulomb blur uniform, it is desirable to consider that the deviation of the pattern density is within 10% when determining the lattice size.

In addition, in the second embodiment, when there is a displacement of a predetermined value or more in the grid-shaped area or a donut pattern exists, the grid-shaped area is divided again into a checkered grid-shaped area. However, as in the first embodiment, even if the donut pattern in the grid-like region is divided in half, the displacement due to the stress acting on the mask is less than a predetermined value, and the tongue problem occurs. If there is no such case, the donut pattern may simply be divided into halves as in the conventional case.

Incidentally, in the first and second embodiments,
There may be a complicated mask pattern, and various reference expressions are required for the displacement at each position. Therefore, in performing these methods, data in which a reference formula such as Formula 1 or Formula 2 for various patterns is tabulated in advance is stored in a storage device so that the data can be read as appropriate. Is desirable.

The first and second embodiments are L / S
It can also be applied to a (line and space) pattern. For example, in a silicon stencil mask having a thickness of 2 μm, an L / S pattern having a width of 0.48 μm has a lattice size of 40 μm or less and a width of 0.40 μm. / S pattern has a lattice size of 20 μm or less and a width of 0.32 micron L
In the / S pattern, the lattice size may be 10 μm or less.

Next, a third embodiment of the present invention will be described. The third embodiment is a pattern design method for the L / S pattern shown in FIG. FIGS. 4A and 4B are plan views showing the first and second complementary masks obtained according to the third embodiment, respectively.

In the third embodiment, the position and shape of the mask hole are determined so that the displacement due to the stress acting on each position of the complementary mask is less than a predetermined value with respect to the L / S pattern. That is, as shown in Formula 1, in the complementary mask, the larger the width of the part that is the support beams at both ends and the shorter the length of the part, the smaller the maximum displacement. Therefore, specifically, FIG.
11B, the space portions in the L / S pattern shown in FIG. 11A are alternately distributed to the first and second complementary masks 4 and 5 to form the mask holes 6.

As described above, in the first and second complementary masks 4 and 5 manufactured according to the third embodiment, since the space portions of the L / S pattern are alternately distributed, each complementary mask 4 and The width of the beam portion in 5 is wider than that in the mask pattern, and the maximum displacement is small. Therefore, the line parts are prevented from contacting each other in the subsequent cleaning process. In addition, the pattern density is the first and second
Since the complementary masks 4 and 5 are uniform, the Coulomb blur is equal.

In the L / S pattern, of course,
There is no need to consider donut, tongue and leaf problems.

Next, a fourth embodiment of the present invention will be described. The fourth embodiment is another pattern design method for the L / S pattern shown in FIG. Figure 5 (a)
And (b) are plan views showing first and second complementary masks obtained by the fourth embodiment, respectively.

Also in the fourth embodiment, the position and shape of the mask hole are determined so that the displacement due to the stress acting on each position of the complementary mask is less than a predetermined value with respect to the L / S pattern. At this time, similarly to the third embodiment, the position and shape of the mask hole can be determined in consideration of, for example, the reference expression of Expression 1. Specifically, FIGS. 5A and 5B
11A, the space portion in the L / S pattern shown in FIG. 11A is divided into upper and lower parts, the upper half is distributed to the first complementary mask 7, and the lower half is distributed to the second complementary mask 8. To form the mask hole 9.

As described above, in the first and second complementary masks 7 and 8 manufactured according to the fourth embodiment, the space portions of the L / S pattern are divided into upper and lower parts and distributed. The length of the beam portion in the masks 7 and 8 is half that in the mask pattern, and the maximum displacement becomes small. Therefore, the line parts are prevented from contacting each other in the subsequent cleaning process. Further, since the pattern density is uniform between the first and second complementary masks 7 and 8, Coulomb blur is equal.

In the third and fourth embodiments, it is first necessary to consider the mask material and shape of the mask pattern and extract the dangerous points where the line portions are likely to come into contact with each other. In comparison with the second embodiment, it takes a long time.

Further, in the first to fourth embodiments, when the pattern of the complementary mask is determined based on the preset reference formula, one opening pattern is distributed to the two complementary masks, and as a result, In the complementary mask, a mask hole having an extremely narrow width may be set. In such cases,
Since the processing becomes difficult, it is desirable to form one distributed mask hole in the other complementary mask. Less than,
Such a method is called boundary processing. Figure 6 is L / S
7A and 7B are plan views showing a part of the pattern.
8A and 8B are plan views showing the first and second complementary masks when the L / S patterns shown in FIG. 6 are distributed based on the reference formula, and FIGS. 8A and 8B are views after the boundary processing, respectively. It is a top view which shows the 1st and 2nd complementary mask.

When the L / S pattern 10 shown in FIG. 6 is distributed, only when it is based on the reference formula, as shown in FIGS. 7A and 7B, the first complementary mask 11 has an extremely wide width. A narrow mask hole 13a may be assigned, and a slightly narrow mask hole 13b may be assigned to the second complementary mask 12. In such a case, as shown in FIGS. 8A and 8B, if the mask hole 13a is set to the second complementary mask 12, the mask hole 13c having a wider width is formed from the mask holes 13a and 13b. Therefore, both complementary masks 11 and 1
From 2, the narrow mask hole is eliminated, and its processing becomes easy. However, also in this case, the two complementary masks 11 and 1
It is desirable that the pattern densities between the two are similar, for example, the deviation is within 10%.

Further, in the fourth embodiment and the like, a technique called multiple exposure can be adopted. FIG. 9 (a)
6B and 6B are plan views showing the first and second complementary masks when multiple exposure is adopted in the L / S pattern shown in FIG. 6, respectively.

When the multiple exposure is adopted, as shown in FIGS. 9A and 9B, the space portion in the L / S pattern shown in FIG. 6 is divided into upper and lower parts, and the first and second complementary masks 14 are formed. , 15, the end shape of the mask hole 16 on the boundary B side is tapered.

By adopting such multiple exposure, even if a slight deviation occurs in the connection of the exposed patterns, the effect of reducing the fluctuation of the line width can be obtained.

The number of complementary masks for one mask pattern is not limited to two and may be three or more. However, if three or more complementary masks are used, the number of exposures increases. Therefore, from the viewpoint of throughput, it is desirable that the number of complementary masks is two.

[0065]

As described in detail above, according to the present invention,
Since the shape of the opening is determined so that the displacement after forming the opening is less than a predetermined value, for example, a cleaning process using pure water,
It is possible to suppress the deformation of the mask even in the mounting process to the holder and the electron beam exposure process in which the temperature rises. Therefore, the yield of the electron beam exposure mask can be improved and the dimensional accuracy thereof can be improved.

Also, by allocating the grid-like areas so as to form a checkerboard pattern, a complicated algorithm is not required, and the design can be made extremely easy.
Further, since the pattern density becomes substantially uniform between the complementary masks, Coulomb blur becomes uniform and the accuracy of the exposed pattern can be improved.

[Brief description of drawings]

FIG. 1 is a flow chart showing a pattern design method according to a first embodiment of the present invention.

FIGS. 2A and 2B are plan views showing first and second complementary masks obtained according to the first embodiment, respectively.

FIG. 3 is a flowchart showing a pattern design method according to a second embodiment of the present invention.

FIGS. 4A and 4B are plan views showing first and second complementary masks obtained according to a third embodiment, respectively.

5A and 5B are plan views showing first and second complementary masks obtained by the fourth embodiment, respectively.

FIG. 6 is a plan view showing a part of an L / S pattern.

7 (a) and 7 (b) are first and second images, respectively, when the L / S pattern shown in FIG. 6 is distributed based on a reference formula.
3 is a plan view showing a complementary mask of FIG.

8A and 8B are plan views showing the first and second complementary masks after the boundary processing, respectively.

9A and 9B are first and second views when multiple exposure is adopted in the L / S pattern shown in FIG. 6, respectively.
3 is a plan view showing a complementary mask of FIG.

10A is a plan view showing a designed mask pattern having a donut problem, and FIG. 10B is a plan view showing a complementary mask to the mask pattern shown in FIG. 10A.

11A is a plan view showing a stencil mask immediately after forming a mask hole, FIG. 11B is a plan view showing an example of a complementary mask to the mask pattern shown in FIG. 11A, and FIG. 11C is another complementary mask. It is a top view which shows an example.

FIG. 12 is a plan view showing an example of a stencil mask.

[Explanation of symbols]

1, 2, 4, 5, 7, 8, 11, 12, 14, 15; Complementary masks 3, 6, 9, 13a, 13b, 13c, 16; Mask hole 10; Mask pattern

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-11-87209 (JP, A) JP-A-11-329948 (JP, A) JP-A-3-64016 (JP, A) JP-A-7- 29793 (JP, A) JP-A-2-166717 (JP, A) JP-A 63-110634 (JP, A) Actual development Shou 57-97941 (JP, U) (58) Fields investigated (Int.Cl. ⁷ , DB name) H01L 21/027 G03F 1/16

Claims (7)

(57) [Claims] 1. A case in which an opening exists only in one grid-like area between adjacent grid-like areas when divided into a plurality of grid-like areas having a certain size, and the opening exists.
A mask for electron beam exposure, characterized in that the child-like region has a checkered pattern . 2. A plurality of grid-shaped regions divided into a grid.
First selection areas that are not adjacent to each other and the first selection area
A first opening formed in at least one of the regions;
Of the areas other than the first selection area of the grid-like area
Divided into at least one region of
A plurality of second selection areas that are smaller than the first selection area;
A second opening formed in any one of the
Area other than the first selection area and the second selection area
Area and the third selection area in which the opening is not formed
An electron beam exposure mask having. 3. The electron beam exposure mask according to claim 1 , wherein the constant dimension is determined in association with the Young's modulus and thickness of the material and the mask pattern to be exposed. 4. A pattern designing method for an electron beam exposure mask, comprising a step of assigning a mask pattern to a plurality of complementary masks, wherein a displacement caused by a stress acting on each position in the complementary mask after forming an opening is It has a step of determining the shape of the opening to be less than a predetermined value, determining the shape of the opening, the mask
The process of partitioning the pattern into a plurality of grid-like regions of a certain size.
And adjacent grid-shaped regions with different complementary masses.
And a step of allocating the mask to the electron beam exposure mask pattern designing method. 5. The mask for electron beam exposure according to claim 4 , wherein the number of the complementary masks is two or more, and the assigned grid-like regions form a checkered pattern in any of the complementary masks. Design method. 6. The step of partitioning the mask pattern,
Design of the electron beam exposure mask according to claim 4 or 5, further comprising the step of determining the dimensions of the grid-like region in relation to Young's modulus and thickness, as well as the mask pattern of the material of said complementary mask Method. 7. The mask pattern is line and
The mask for electron beam exposure according to claim 4 , wherein the mask is a space-like pattern, and the step of determining the shape of the opening is performed in association with the width, length and thickness of the line portion. Design method.